* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Math F412: Homework 7 Solutions March 20, 2013 1. Suppose V is
Vector space wikipedia , lookup
Gaussian elimination wikipedia , lookup
Singular-value decomposition wikipedia , lookup
Cayley–Hamilton theorem wikipedia , lookup
Orthogonal matrix wikipedia , lookup
Matrix multiplication wikipedia , lookup
Eigenvalues and eigenvectors wikipedia , lookup
Jordan normal form wikipedia , lookup
Covariance and contravariance of vectors wikipedia , lookup
Matrix calculus wikipedia , lookup
Perron–Frobenius theorem wikipedia , lookup
Math F412: Homework 7 Solutions March 20, 2013 1. Suppose V is a finite-dimensional inner-product space with inner product ⟨⋅, ⋅⟩ and that T ∶ V → V is symmetric. Show that T has no complex eigenvalues. Hint: Let W be the complex vector space of vectors of the form a + ib where a, b ∈ V . You need not show that this is a vector space. We extend T to a map T ∶ W → W by T(a + ib) = Ta + iTb. It’s easy to see that this map is complex linear; don’t show this. If the characteristic equation det(T − λI) has a complex root λ, then there is a vector z = x + i y in W such that T(x + i y) = λ(x + i y); there is a little work to be done here, but don’t do it either unless you are looking for some extra fun. Extend the inner product to complex vectors in W by ⟨a + ib, w⟩ = ⟨a, w⟩ + i ⟨b, w⟩ and ⟨w, a + ib⟩ = ⟨w, a⟩ + i ⟨w, b⟩ . Now show 1. Tz = λz if z is an eigenvector with eigenvalue λ. 2. < Tz, w >= ⟨z, Tw⟩ for all complex vectors z and w. From these two ingredients you can show that any eigenvalue of T must be real by looking at ⟨Tz, z⟩. Solution: We suppose z = x + i y is a complex eigenvector of T with eigenvalue λ. Note that T(z) = T(x − i y) = Tx − iTy = Tx + iTy = T(x + i y) = λz = λz. Now suppose v, w ∈ W. Then v = a + ib and w = c + id for some a, b, c, d ∈ V . Then ⟨Tv, w⟩ = ⟨T(a + ib), c + id⟩ = ⟨Ta + iTb, c + id⟩ = ⟨Ta, c⟩ + i ⟨Ta, d⟩ + i ⟨Tb, c⟩ − ⟨Tb, d⟩ . Using the symmetry of T acting on V we find ⟨Tv, w⟩ = ⟨a, Tc⟩ + i ⟨a, Td⟩ + i ⟨b, Tc⟩ − ⟨b, Td⟩ = ⟨a + ib, Tc + iTd⟩ = ⟨v, Tw⟩ . Now consider ⟨Tz, z⟩ = ⟨λz, z⟩ = λ < z, z > . But ⟨Tz, z⟩ = ⟨z, Tz⟩ = ⟨z, λz⟩ = λ ⟨z, z⟩ . Math F412: Homework 7 Solutions March 20, 2013 Hence λ ⟨z, z⟩ = λ ⟨z, z⟩ . One readily verifies that ⟨z, z⟩ = ⟨x, x⟩ + ⟨y, y⟩ ≠= 0. So λ=λ and λ is real. 2. Let A and B be n × n matrices. Show tr AB = tr BA. Conclude that if T ∶ V → V is a linear map between finite dimensional vector spaces that tr T is well-defined. Solution: Let A = [A i j] and B = [B i j] so that n AB i j = ∑ A ik B k j . k=1 Then n n n n n tr AB = ∑ AB ii = ∑ ∑ A ik B ki = ∑ ∑ B ki A ik = tr BA. i=1 i=1 k=1 k=1 i=1 Now if A is any n × n matrix and P is any invertible n × n matrix, tr P −1 AP = tr APP −1 = tr A. Now suppose T ∶ V → V is linear and V is finite dimensional. Let B1 and B2 be two bases on V , and let P be the change of basis matrix introduced in class. Then TB2 = P −1 TB1 P. But then tr TB2 = tr P −1 TB1 P = tr TB1 PP 1 = tr TB1 . Hence the trace of TB is independent of the choice of basis B and can be used to define tr T. 3. Let B be a basis for Tp M. a) Show that IB SB = IIB , where S = S p is the shape operator on Tp M. b) Explain why IB is an invertible 2 × 2 matrix. c) Show that SB = IB−1 IIB . d) Conclude that K= and H= ln − m2 EG − F 2 Gl + En − 2Fm . 2(EG − F 2 ) 2 Math F412: Homework 7 Solutions March 20, 2013 Solution, part a: Let B = {x1 , x2 } be a basis for Tp M. Let IIB = [II i j ], IB = [I i j ], and SB = [S i j ]. Then II i j = II(x i , x j ) = ⟨Sx i , x j ⟩ = ⟨x i , Sx j ⟩ = ⟨x i , x k S k j ⟩ = ⟨x i , x k ⟩ S k j = II ik S k j . Hence IIB = IB SB . Solution, part b: Suppose to the contrary that IB is not invertible. Then there is a nonzero vector v such that II i j v j = 0 for each i. So v i II i j v j = 0 Let w = v1 x1 + v2 x2 so that wB = v. Then ⟨w, w⟩ = v t IIB v = 0. Hence w = 0 and therefore v = 0. This is a contradiction. Solution, part c: This follows immediately from parts (a) and (b). Solution, part d: Using a chart to determine the usual basis xu and xv for Tp M we write IB = [ We see that E F ] F G IIB = [ l m ]. m n K = det S = (det IB )−1 det IIB = (ln − m2 )/(EG − F 2 ). Also IB−1 = So IB−1 IIB = 1 G −F [ ]. 2 EG − F −F E 1 Gl − Fm Gm − Fn [ ]. 2 −Fm + El −Fm + En EG − F Consequently H= GL + En − 2Fm 1 tr IB−1 IIB = 2 2(EG − F 2 ) 4. Write a Maple procedure UnitNormal that takes an expression for a chart x and the name of two coordinate variables (e.g. u and v) and returns a simplified expression for the unit normal in terms of u and v. Verify your procedure works by computing the unit normal of the helicoid x(u, v) = (v cos u, v sin u, bv). (You have already computed this normal on a previous homework.) 3 Math F412: Homework 7 Solutions March 20, 2013 5. Write a Maple procedure FundamentalForms that takes an expression for a chart x and the name of two coordinate variables (e.g. u and v) and returns two 2 × 2 matrices, the matrices of the first and second fundamental forms computed with respect to the basis {xu , xv }. 6. Compute E, F, and G, as well as l, m, and n for the chart x(u, v) = (cos(u) sin(v), sin(u) sin(v), cos(v)). You may use Maple to assist you in your computation. Explain why your results for F and G make intuitive sense. Solution: See worksheet for the computation that gives F = 0 and G = 1. These make sense because the lines of latitude and longitude are orthogonal, and because we parameterized lines of longitude with a unit speed curve. 7. Let x be a chart with domain D into the surface M. Let α̃ ∶ [0, T] → D be a curve in D, so α̃(t) = (u(t), v(t)) for some functions u and v. Let α(t) = x(α(t)), so α is a curve in M. Show that L(α) = ∫ T 0 E(α̃(t))(u′ (t))2 + 2 ∗ F(α̃(t))u′ (t)v ′ (t) + G(α̃(t))(v ′ (t))2 dt. This shows us that E, F, and G can be used to compute the lengths of curves in local coordinates. Let x be the chart in problem 6. Let α̃(t) = (t, π/4) where −π < t < π. Use the formula developed above and the values of E, F, and G computed in problem 6 to compute the length of α. Explain why the result of this computation makes intuitive sense. Solution: Note that α ′ (t) = xu (α(t))u′ (t) + xv (α(t))v ′ (t). Hence ∣α ′ (t)∣2 = ⟨xu (α(t))u′ (t) + xv (α(t))v ′ (t), xu (α(t))u′ (t) + xv (α(t))v ′ (t)⟩ = E(u′ )2 +2Fu ′ v ′ +G(v ′ )2 . Hence L(α) = ∫ T 0 ∣α ′ (t)∣ dt = ∫ T 0 √ E(α̃(t))(u′ (t))2 + 2 ∗ F(α̃(t))u′ (t)v ′ (t) + G(α̃(t))(v ′ (t))2 dt. For the sphere example we have E = sin2 v, F = 0, G = 1, u′ (t) = 1, and v ′ (t) = 0. So √ π√ 2 2 L=∫ sin π/4 dt = 2π sin π/4 = 2π . 2 −π This result makes sense because the curve on the sphere is the line of latitude for angle π/4 which is a circle of radius cos π/4. 8. Write a Maple procedure that computes the Gaussian and mean curvatures of a surface. Then compute the Gaussian and mean curvatures of the helicoid. Solution: See worksheet. 4 Math F412: Homework 7 Solutions March 20, 2013 9. Write a Maple procedure Shape that takes an expression for a chart x and the name of two coordinate variables (e.g. u and v) and returns the matrix of the shape operator with respect to the basis xu and xv . Solution: See worksheet. 10. Use the program you wrote in the previous problem to re-do Exercise 2.2.16 in Oprea. Solution: See worksheet. 5